L1/L5 SBAS MOPS to Support Multiple Constellations

1
L1/L5 SBAS MOPS to Support Multiple Constellations

• Todd Walter, Juan Blanch,
• and Per Enge
• Stanford University
• http//waas.stanford.edu

2
Background

• At IWG 20 in Madrid it was agreed that the L5
  MOPS should support 4 constellations with 90
  active corrections
• At IWG 21 in Stanford it was agreed that there
  was interest in providing a better level of
  service than LPV-200 (e.g. CAT-II)
• Details of service are uncertain

3
Discussion

• At IWG 24 in Toulouse some concerns were raised
  over Stanfords proposed L5 MOPS (called ICD2)
• Longer interval between clock updates may require
  larger bounding term
• Use of change indicators rather than full DFREs
• Ability to alert more than 7 SV due to single
  event
• Effect of message loss for this alert message
• Delayed broadcast of DFREI raises effective value
  (called border effect)
• Alternate proposal (called ICD1) followed L1 MOPS
  more closely
• Requires two MT6s to support more than 51 total
  corrections

4
Proposed ICD2 Components

• Expanded PRN mask
• Removal of FCs
• Alert message with DFRECIs
• Single satellite correction message
• Smaller quantization errors
• Updatable DFRE table
• Support for more SBAS orbit types

5
Expanded PRN Mask

• 37 PRNs per constellation
• GPS, GLONASS, Galileo, Compass
• 39 PRNs available for SBAS GEOs
• 23 spare PRNs for QZSS, IRNSS etc.
• More possible if fewer than 37 given to each
  constellation
• Fully compatible with existing L1 MOPS and other
  proposed L5 changes (ICD1 ICD2)

6
Need for 63 GPS PRNs?

• GPS ICD has 63 defined PRNs
• The definition table for PRN codes also has 210
  slots and assigns the first 63 to GPS
• SBAS MOPS (229D) already assigns slots 38-61 to
  GLONASS
• No need to match tables 1-to-1
• Can have a mapping from one to the other
• Do we need to support 63 GPS PRNs?
• Will other constellations want the same?
• Could no longer fit SBAS PRN mask into a single
  message
• Unlikely to have more than 37 active PRNs

7
Removal of Fast Corrections

• Fast correction originally intended to support
  Selective Availability (SA)
• No longer needed as satellite atomic clocks are
  extremely predictable over time periods of
  interest (lt 400 secs)
• Velocity ltlt 1 cm/sec
• Acceleration is immeasurably small over a few
  minutes
• More details later in presentation

8
New Alert Message

• Allows more than 51 corrections
• By sending DFRE Change Indicators instead of full
  DFREIs
• Alternate methods to achieve gt 51
• Longer updates between SV corrections
• Increase bit rate (e.g. use Q. channel)
• Not require DFRE update every 6 sec
• ICD1 achieves this goal with two type 6 messages
• Extends LTC update interval to 240 sec

9
Single Satellite Correction Message

• MTs 25 28 each provide updates for two
  satellites
• Loss of either message affects both SVs
• Instead make one message contain the LTC and
  covariance for one SV
• Cleaner connection between the two
• Save some of the SV ID bits
• Still an improvement even if fast correction not
  eliminated
• Compatible with ICD1 ICD2

10
Smaller Quantization

• 12.5 cm resolution is a noticeable contributor to
  WAAS inaccuracy
• Clock dynamic range originally set due to SA
  can be much smaller
• Reducing dynamic range and quantization error
  improves accuracy for both ICD1 and ICD2
• For ICD1 would also recommend eliminating RRC by
  differencing adjacent fast corrections

11
Updatable DFRE Table

• Broadcasting the DFRE table allows the DFRE
  quantization to be optimized to the service
• Ensure use of all DFREIs
• Minimize quantization penalty
• User applies maximum value if they have not yet
  received the table
• Changes would be very rare and tightly controlled
  (perhaps MT0 in between)
• Compatible with ICD1 ICD2

12
SBAS SV Orbit Messages

• Specifies augmented Keplerian elements
• Allows for more than just geostationary and
  near-geostationary orbits
• Better accuracy
• Better long-term performance for almanacs
• Retain existing GEO messages
• This is an additional/alternate message
• Compatible with ICD1 or ICD2

13
SV Clock Errors Rate
Data from March 24-30, 2013
14
SV Clock Error Acceleration
Data from March 24-30, 2013
15
Linear Behavior of SV Clocks
Data from March 24, 2013
16
Linear Prediction
Clock Data
MOPS Corrections (12.5 cm resolution)
Extrapolated Fit
ICD2 Correction (3 cm resolution)
Linear Fit
17
Prediction Accuracy
1-s Values No Quant. L1 MOPS (6 s) ICD2
Average 1.8 cm 0.03 cm/s 4.5 cm 0.09 cm/s 1.8 cm 0.03 cm/s
Worst SV overbound 6 cm 0.17 cm/s 6.5 cm 1 cm/s 6.1 cm 0.17 cm/s
18
Pseudorange Overbound

• L1 MOPS with current quantization and 6 sec FC
  updates requires a PR overbound of gt 12.5 cm
  after 6 sec
• ICD1 without quantization error and 6 sec FC
  requires a PR overbound of gt 7 cm after 6 sec
• ICD2 requires a PR overbound of gt 12.2 cm after
  36 sec
• Could set acceleration bound to zero and include
  small linear growth term
• Slower clock broadcast for 1 constellation should
  lead to less than than a 1 m difference in VPL

19
Border Effect

• For one constellation, no more than 3 DFREIs
  require update within a 36 second interval gt99.9
  of the time
• For two constellations, no more than 5 DFREIs
  require update within a 60 second interval gt99.9
  of the time
• Nominally, there should be no border effect for
  one or two constellations
• Even for three or four constellations, the border
  effect should only delay a small number of DFREIs
  by order 6-18 seconds

20
Nominal DFREI Changes per Time Interval
Predicted WAAS DFREI changes due to geometry for
31 SV GPS constellation excluding changes to NM
21
Change Indicator Interpretation 1

• A no change (NC) indication points back to the
  full DFREI in the satellite correction
• Changes may need to be indicated until 2 or more
  correction message broadcasts
• Or could use degradation parameters

I
Integrity message
Satellite correction
NC 1
DFREI 6
DFREI 1 7
DFREI 1 7
DFREI 1 7
DFREI 1 7
NC 1 or DFREI 1 7
DFREI 1 7
DFREI 1 7
22
Change Indicator Interpretation 2

• A no change indication points to the most
  recently received DFREI in the last 18 sec
• Three successive DFREI updates are sufficient to
  fully update DFREI
• Unable to determine DFREI if three prior
  integrity messages are lost

I
Integrity message
Satellite correction
NC 1
DFREI 6
DFREI 1 7
DFREI 1 7
DFREI 1 7
DFREI 1 7
NC 1
NC 1
NC 1
23
Interpretation 2

• If a user has not received a full DFREI update
  for a satellite in the last 18 seconds
• The satellite must be set to NM unless it is
  already DNU
• It remains in that state until a full DFREI value
  is obtained
• unless the DFRECI sets it to NM or DNU
• The loss of three successive integrity messages
  could cause all satellites to go NM
• Three independent message outages is very rare (lt
  10-9)
• A single outage would need to last 13 seconds or
  longer
• Should also be rare
• Recovery may be slow as will need to wait for
  satellite corrections to be broadcast
• Up to 120 seconds for all satellites
• However, the border effect on availability should
  be negligible even for four constellations

24
Discussion Point 1

• The decision to support more than 51 active
  corrections can be revisited
• Did not want the MOPS to constrain future more
  demanding operations
• Additional constellations support lower
  protection levels
• Requirements for such operations have not yet
  been determined

25
99.5 VPL
1 Constellation
2 Constellations
3 Constellations
4 Constellations
26
99.5 HPL
1 Constellation
2 Constellations
3 Constellations
4 Constellations
27
HPL VPL as a Function of Number of Satellites
24 SVs
47 SVs
74 SVs
101 SVs
28
Discussion Point 2

• Fast Corrections are primarily only meaningful
  for a single constellation
• For two constellations they are slowed to 33
  second intervals or longer
• For three or more constellations they are on par
  with the long term corrections and limit ability
  to fit within bandwidth
• However legacy ground systems are set up to
  generate Fast Corrections
• May be lower initial cost to retain them

29
Recommendations

• The remaining ICD2 components can be implemented
  with either approach
• Recommend adopting
• Expanded PRN mask
• Combining Type 25 and Type 28 info into a single
  satellite correction message
• Reducing quantization errors
• And eliminating RRC from adjacent FCs
• Updatable DFRE table
• Support for more SBAS orbit types as an
  additional message

30
(No Transcript)